Some operations on complement bitsets are wrong about red HOT 3 OPEN

The length of the bitset should be taken into account.

>> c1
==  make bitset! #{FFFF}      ;-- "11111111.11111111"
>> c2
== make bitset! [not #{7E}]   ;-- "10000001"
>> length? c1
== 16
>> length? c2
== 8
>> c3: c1 and c2
== make bitset! [not #{7E}]   ;-- "10000001"
>> length? c3
== 8

c1 and c2 is correct.

from red.

No it's not correct. Length is just an internal implementation detail. Implementation should not drive the design. Design should drive the implementation.

from red.

General rules for bitsets should be:

1. Result length should equal the longest operand's length, and may be zero:

   >> (charset []) or (charset [not]) 
   == make bitset! [not #{FF}]     ;) length was extended leading to wrong result! expected [not #{}]
   

2. Negation flag may be freely set by binary operations depending on whether the trail should be set or not, e.g.:

   >> (charset [not]) and (charset [])
   == make bitset! [not #{}]       ;) expected empty set, got full set!
   >> (charset [not]) and (charset [not]) 
   == make bitset! #{}             ;) expected full set, got empty set!
   

Current builds of Red, R3 and Ren-C I have all fail the bitset math. @Oldes @hostilefork FYI.

from red.